请提出如何在oracle中实现以下o / p？

Question

输入：

insert into example (fileno,type_cd,date_tm,be_nm) values('1','mismatch','match','mismatch');
insert into example (fileno,type_cd,date_tm,be_nm) values('2','match','mismatch','mismatch');
insert into example (fileno,type_cd,date_tm,be_nm) values('3','match','mismatch','match');
insert into example (fileno,type_cd,date_tm,be_nm) values('4','match','mismatch','mismatch');

Answer 1

您可以取消设置初始数据：

select *
from example
unpivot (match for column_name in
  (type_cd as 'type_cd', date_tm as 'date_tm', be_nm as 'be_nm'));

    FILENO COLUMN_ MATCH   
---------- ------- --------
         1 type_cd mismatch
         1 date_tm match   
         1 be_nm   mismatch
         2 type_cd match   
         2 date_tm mismatch
         2 be_nm   mismatch
...

然后将其旋转：

select *
from (
  select column_name, match
  from example
  unpivot (match for column_name in
    (type_cd as 'type_cd', date_tm as 'date_tm', be_nm as 'be_nm'))
)
pivot (count(*) for (match) in ('match' as match, 'mismatch' as mismatch));

COLUMN_NAME      MATCH   MISMATCH
----------- ---------- ----------
type_cd              3          2
date_tm              1          4
be_nm                2          3

或者稍微减少打字，但也许也不太清楚：

select *
from (select type_cd, date_tm, be_nm from example)
unpivot (match for column_name in
  (type_cd as 'type_cd', date_tm as 'date_tm', be_nm as 'be_nm'))
pivot (count(*) for (match) in ('match' as match, 'mismatch' as mismatch));

COLUMN_NAME      MATCH   MISMATCH
----------- ---------- ----------
type_cd              3          2
date_tm              1          4
be_nm                2          3

我假设您在输入中错过了5 /错配/错配/匹配的行，因为没有它，您将不会获得这些总数。

---

或者有一种蛮力的方法，它涉及条件聚集并将多个查询的结果结合在一起，这需要多次击中表：

select 'type_cd' as column_name,
  count(case when type_cd = 'match' then 1 end) as match, 
  count(case when type_cd = 'mismatch' then 1 end) as mismatch
from example
union all
select 'date_tm' as column_name,
  count(case when date_tm = 'match' then 1 end) as match, 
  count(case when date_tm = 'mismatch' then 1 end) as mismatch
from example
union all
select 'be_nm' as column_name,
  count(case when be_nm = 'match' then 1 end) as match, 
  count(case when be_nm = 'mismatch' then 1 end) as mismatch
from example;

COLUMN_NAME      MATCH   MISMATCH
----------- ---------- ----------
type_cd              3          2
date_tm              1          4
be_nm                2          3

无论是性能还是可维护性，unpivot / pivot方法都可以更好地扩展。